@memoarchitect/tools 0.4.4

package/packages/tools/src/grammar/memo-sysml.langium

@@ -0,0 +1,450 @@
+// =============================================================================
+// MEMO SysML v2 Grammar — Langium definition
+//
+// Covers the SysML v2 subset needed for MEMO systems engineering modeling:
+//   package, part def, requirement def, action def, item def, port def,
+//   interface def, connection def, attribute def, enum def,
+//   viewpoint def, view def, import, specialization (:>),
+//   typed connections, doc comments
+//
+// Behavior constructs (Phase 5):
+//   action def with in/out/inout parameters, action usage (composite),
+//   flow connection usage, succession usage, allocate usage, item def
+//
+// Design notes:
+//   - DOC_COMMENT is a non-hidden terminal that captures `doc /* ... */`
+//     as a single token, preventing the hidden ML_COMMENT from consuming it
+//   - FeatureName allows 'part' keyword as identifier in end/connection contexts
+//     (SysML v2 uses `end part[0..*]` in connection defs)
+//   - QualifiedName is a data type rule returning string; cross-references
+//     are resolved programmatically in the semantic model layer
+// =============================================================================
+
+grammar MemoSysML
+
+// ─── Entry point ─────────────────────────────────────────────────────────────
+
+entry Model:
+    members+=NamespaceMember*;
+
+NamespaceMember:
+    PackageDeclaration | ImportDeclaration;
+
+// ─── Package ─────────────────────────────────────────────────────────────────
+
+// Standard SysML v2: a package DECLARATION takes a single-identifier name.
+// Qualified names (`::`) are for references only; deep namespaces are expressed by
+// NESTING packages (`package memo { package arch { … } }`). The model builder
+// reconstructs the full FQN from the nesting (builder.ts: parentPackage::pkg.name).
+PackageDeclaration:
+    isLibrary?='library'? 'package' name=ID '{'
+        members+=PackageMember*
+    '}';
+
+PackageMember:
+    PackageDeclaration
+    | AliasDeclaration
+    | ImportDeclaration
+    | DefinitionMember
+    | UsageMember
+    | ConnectionUsage
+    | AllocateUsage
+    | MetadataApplication
+    | DocComment;
+
+// Namespace indirection used by the canonical memo::* facade. Each source file
+// owns a unique package for reliable SysIDE indexing; aliases reconstruct the
+// user-facing hierarchy without reopening packages across documents.
+AliasDeclaration:
+    'alias' name=ID 'for' target=QualifiedName ';';
+
+// ─── Import ──────────────────────────────────────────────────────────────────
+
+ImportDeclaration:
+    visibility=('private' | 'public')? 'import' path=ImportPath ';';
+
+// Separate rule to handle wildcard imports (e.g. `import MEMO_Ontology_Medical::*`)
+// without LL parser ambiguity on `::` lookahead in QualifiedName
+ImportPath returns string:
+    ID ('::' (ID | '*'))*;
+
+// ─── Definitions ─────────────────────────────────────────────────────────────
+
+DefinitionMember:
+    PartDefinition
+    | RequirementDefinition
+    | ActionDefinition
+    | ItemDefinition
+    | PortDefinition
+    | InterfaceDefinition
+    | ConnectionDefinition
+    | AttributeDefinition
+    | ConstraintDefinition
+    | EnumDefinition
+    | ViewpointDefinition
+    | ViewDefinition
+    | MetadataDefinition;
+
+// SysML v2 semantic-metadata definition — the standard, portable mechanism for
+// extending the ontology with domain semantics. Applied to elements via `@Name`.
+MetadataDefinition:
+    'metadata' 'def' name=ID specialization=Specialization?
+    ('{' body+=DefinitionBodyMember* '}' | ';');
+
+// Metadata application: `@AnnotationType;` or `@AnnotationType { redefinitions }`
+// applies a metadata def to the enclosing element (standard SysML v2 `@` annotation).
+MetadataApplication:
+    '@' type=QualifiedName ('{' body+=UsageBodyMember* '}' | ';');
+
+PartDefinition:
+    isAbstract?='abstract'? 'part' 'def' name=ID specialization=Specialization? (',' additionalSpecializations+=QualifiedName)*
+    ('{' body+=DefinitionBodyMember* '}' | ';');
+
+RequirementDefinition:
+    isAbstract?='abstract'? 'requirement' 'def' name=ID specialization=Specialization? (',' additionalSpecializations+=QualifiedName)*
+    ('{' body+=DefinitionBodyMember* '}' | ';');
+
+ActionDefinition:
+    isAbstract?='abstract'? behaviorKind=('action' | 'operator' | 'function') 'def' name=ID specialization=Specialization? (',' additionalSpecializations+=QualifiedName)*
+    ('{' body+=ActionDefBodyMember* '}' | ';');
+
+ItemDefinition:
+    isAbstract?='abstract'? 'item' 'def' name=ID specialization=Specialization? (',' additionalSpecializations+=QualifiedName)*
+    ('{' body+=DefinitionBodyMember* '}' | ';');
+
+PortDefinition:
+    isAbstract?='abstract'? 'port' 'def' name=ID specialization=Specialization? (',' additionalSpecializations+=QualifiedName)*
+    ('{' body+=DefinitionBodyMember* '}' | ';');
+
+InterfaceDefinition:
+    isAbstract?='abstract'? 'interface' 'def' name=ID specialization=Specialization? (',' additionalSpecializations+=QualifiedName)*
+    ('{' body+=DefinitionBodyMember* '}' | ';');
+
+ConnectionDefinition:
+    isAbstract?='abstract'? 'connection' 'def' name=ID specialization=Specialization? (',' additionalSpecializations+=QualifiedName)*
+    ('{' body+=ConnectionBodyMember* '}' | ';');
+
+AttributeDefinition:
+    isAbstract?='abstract'? 'attribute' 'def' name=ID specialization=Specialization? (',' additionalSpecializations+=QualifiedName)*
+    ('{' body+=DefinitionBodyMember* '}' | ';');
+
+ConstraintDefinition:
+    isAbstract?='abstract'? 'constraint' 'def' name=ID specialization=Specialization? (',' additionalSpecializations+=QualifiedName)*
+    ('{' body+=DefinitionBodyMember* '}' | ';');
+
+EnumDefinition:
+    'enum' 'def' name=ID '{'
+        literals+=EnumLiteral*
+    '}';
+
+ViewpointDefinition:
+    isAbstract?='abstract'? 'viewpoint' 'def' name=ID specialization=Specialization? (',' additionalSpecializations+=QualifiedName)*
+    ('{' body+=DefinitionBodyMember* '}' | ';');
+
+ViewDefinition:
+    isAbstract?='abstract'? 'view' 'def' name=ID specialization=Specialization? (',' additionalSpecializations+=QualifiedName)*
+    ('{' body+=DefinitionBodyMember* '}' | ';');
+
+// ─── Connection definition body ─────────────────────────────────────────────
+
+ConnectionBodyMember:
+    EndDeclaration | DefinitionBodyMember;
+
+EndDeclaration:
+    'end' name=FeatureName (':' type=QualifiedName)? multiplicity=Multiplicity? ';';
+
+Multiplicity:
+    '[' lower=INT '..' (upper=INT | unbounded?='*') ']'
+    | '[' unbounded?='*' ']'
+    | '[' exact=INT ']';
+
+// ─── Enum literal ────────────────────────────────────────────────────────────
+
+EnumLiteral:
+    'enum' name=EnumLiteralName ';';
+
+EnumLiteralName returns string:
+    ID | 'in' | 'out' | 'inout' | 'part' | 'requirement' | 'action' | 'item' | 'port'
+    | 'interface' | 'connection' | 'attribute' | 'constraint' | 'enum' | 'view' | 'viewpoint'
+    | 'ref' | 'import' | 'package' | 'flow' | 'first' | 'allocate' | 'library' | 'private' | 'public';
+
+// ─── Usages (instances) ─────────────────────────────────────────────────────
+
+UsageMember:
+    PartUsage
+    | RequirementUsage
+    | ActionUsage
+    | PortUsage
+    | ViewUsage;
+
+PartUsage:
+    'part' name=ID ':' type=QualifiedName
+    ('{' body+=UsageBodyMember* '}' | ';');
+
+RequirementUsage:
+    'requirement' name=ID ':' type=QualifiedName
+    ('{' body+=UsageBodyMember* '}' | ';');
+
+// ActionUsage supports both typed (action name : Type;) and composite
+// (action name { nested-actions, flows, successions }) forms.
+ActionUsage:
+    behaviorKind=('action' | 'operator' | 'function') name=ID (':' type=QualifiedName)?
+    ('{' body+=ActionUsageBodyMember* '}' | ';');
+
+PortUsage:
+    direction=('in' | 'out' | 'inout')? 'port' isConjugated?='~'? name=ID ':' type=QualifiedName
+    ('{' body+=UsageBodyMember* '}' | ';');
+
+ViewUsage:
+    'view' name=ID (':' type=QualifiedName)?
+    ('{' body+=ViewBodyMember* '}' | ';');
+
+ViewBodyMember:
+    UsageBodyMember | ExposeMember;
+
+ExposeMember:
+    'expose' path=ExposePath ';';
+
+ExposePath returns string:
+    ID (('::' | '.') ID)* ('::' ('*' | '**'))?;
+
+// ─── Connection usage (typed relationship instances) ─────────────────────────
+// Syntax: connection : Type connect end1 ::> ref1 to end2 ::> ref2 ;
+
+ConnectionUsage:
+    'connection' (':' type=QualifiedName)?
+    'connect' source=ConnectionEnd 'to' target=ConnectionEnd ';';
+
+ConnectionEnd:
+    endName=FeatureName '::>' ref=ReferenceName;
+
+// ─── Behavior: Action definition body ───────────────────────────────────────
+// Action defs can have in/out/inout parameters plus regular body members
+
+ActionDefBodyMember:
+    ActionParameterMember | DefinitionBodyMember;
+
+ActionParameterMember:
+    direction=('in' | 'out' | 'inout') name=ID ':' type=QualifiedName ';';
+
+// ─── Behavior: Action usage body ────────────────────────────────────────────
+// Composite actions contain nested actions, flow connections, successions
+
+ActionUsageBodyMember:
+    ActionUsage | ControlNodeUsage | FlowConnectionUsage | SuccessionUsage | UsageBodyMember;
+
+// ─── Behavior: Control nodes (fork / join) ──────────────────────────────────
+// SysML v2 activity control nodes. A `fork` splits one incoming control flow
+// into concurrent outgoing flows; a `join` synchronizes concurrent flows back
+// into a single flow. They carry no parameters — ordering and the branches
+// they split/merge are expressed by the surrounding succession usages that
+// reference the node by name.
+//   fork parallelPrep;
+//   first verify then parallelPrep;
+//   first parallelPrep then prime;
+//   first parallelPrep then selfTest;
+ControlNodeUsage:
+    controlKind=('fork' | 'join') name=ID ';';
+
+// ─── Behavior: Flow connection usage ────────────────────────────────────────
+// Syntax: flow of ItemType from source.port to target.port ;
+
+FlowConnectionUsage:
+    'flow' 'of' itemType=QualifiedName 'from' source=FlowEnd 'to' target=FlowEnd ';';
+
+FlowEnd:
+    ref=DottedName;
+
+// Dot-separated qualified name for flow endpoints (e.g. receive.prescription)
+DottedName returns string:
+    ID ('.' ID)*;
+
+// ─── Behavior: Succession usage ─────────────────────────────────────────────
+// Syntax: first start then action1 then action2 then done ;
+
+SuccessionUsage:
+    'first' steps+=SuccessionStep ('then' steps+=SuccessionStep)+ ';';
+
+SuccessionStep:
+    ref=SuccessionRef;
+
+SuccessionRef returns string:
+    'start' | 'done' | ID;
+
+// ─── Behavior: Allocate usage ───────────────────────────────────────────────
+// Syntax: allocate actionUsage to partUsage ;
+
+AllocateUsage:
+    'allocate' source=ReferenceName 'to' target=ReferenceName ';';
+
+// Element references may be package-qualified (`a::b`) or feature-qualified
+// (`action.step`). Both forms occur in standard SysML relationship ends.
+ReferenceName returns string:
+    ID (('::' | '.') ID)*;
+
+// ─── Body members ────────────────────────────────────────────────────────────
+
+DefinitionBodyMember:
+    AttributeMember | PartMember | PortUsage | ViewUsage | RefMember | RequireConstraintMember | MetadataApplication | DocComment;
+
+// ─── Constraint expression (KerML boolean subset — ADR-1-18) ─────────────────
+// Authored inside constraint def / requirement def bodies as:
+//   require constraint { <boolean expr> }
+//   assert  constraint { <boolean expr> }
+// Precedence (loosest→tightest): or < and < cmp < add < mul < not < postfix(->) < primary.
+
+RequireConstraintMember:
+    kind=('require' | 'assert')? 'constraint' '{' expression=OrExpr '}';
+
+OrExpr infers ConstraintExpr:
+    AndExpr ({infer BinaryExpr.left=current} op='or' right=AndExpr)*;
+
+AndExpr infers ConstraintExpr:
+    CmpExpr ({infer BinaryExpr.left=current} op='and' right=CmpExpr)*;
+
+CmpExpr infers ConstraintExpr:
+    AddExpr ({infer BinaryExpr.left=current} op=('==' | '!=' | '>=' | '<=' | '>' | '<') right=AddExpr)?;
+
+AddExpr infers ConstraintExpr:
+    MulExpr ({infer BinaryExpr.left=current} op=('+' | '-') right=MulExpr)*;
+
+MulExpr infers ConstraintExpr:
+    UnaryExpr ({infer BinaryExpr.left=current} op=('*' | '/') right=UnaryExpr)*;
+
+UnaryExpr infers ConstraintExpr:
+    {infer UnaryExpr} op='not' operand=UnaryExpr
+    | PostfixExpr;
+
+PostfixExpr infers ConstraintExpr:
+    PrimaryExpr ({infer CollectionOp.target=current} '->' op=ID '(' (argument=OrExpr)? ')')*;
+
+PrimaryExpr infers ConstraintExpr:
+      {infer LiteralExpr} intValue=INT
+    | {infer LiteralExpr} strValue=STRING
+    | {infer LiteralExpr} boolValue=('true' | 'false')
+    | {infer FeatureChain} segments+=ID ('.' segments+=ID)*
+    | '(' OrExpr ')';
+
+// Standard SysML v2 requires a redefinition operator (`redefines` / `:>>`) to bind a
+// value — the bare `name = value;` shorthand is non-standard and is not accepted.
+// Use `attribute redefines name = value;` (the form the serializer emits).
+UsageBodyMember:
+    AttributeMember | PartMember | PortUsage | ViewUsage | MetadataApplication | DocComment;
+
+// ─── Attribute member ────────────────────────────────────────────────────────
+// Two forms:
+//   attribute name : Type ;              (type declaration)
+//   attribute redefines name = value ;   (value assignment / redefinition)
+
+AttributeMember:
+    'attribute' redefines?=('redefines' | ':>>')? name=ID
+    (':' type=QualifiedName multiplicity=Multiplicity? ';' | '=' value=AttributeValue ';');
+
+AttributeValue:
+    StringValue | RealValue | IntValue | BooleanValue | SetLiteral | EnumValue;
+
+StringValue:
+    value=STRING;
+
+// Real (rational) literal, e.g. 20.0 — must precede IntValue so the lexer
+// prefers REAL over INT for the integral part.
+RealValue:
+    value=REAL;
+
+IntValue:
+    value=INT;
+
+BooleanValue:
+    value=('true' | 'false');
+
+SetLiteral:
+    ('{' elements+=SetElement (',' elements+=SetElement)* '}'
+    | '(' elements+=SetElement (',' elements+=SetElement)* ')');
+
+SetElement:
+    value=QualifiedName | stringValue=STRING;
+
+// Enum literal reference: e.g. SeverityLevel::Catastrophic
+EnumValue:
+    enumRef=QualifiedName;
+
+// ─── Nested part member ──────────────────────────────────────────────────────
+// Three forms:
+//   part subsystems : Subsystem[0..*];        (typed nested part)
+//   part requirement = reqSingleMode;          (feature value binding to a reference)
+//   part protectedAsset :> therapyAsset;       (subsetting an existing element)
+// The binding forms carry the source/target ends of semantic *Link instances,
+// which the model builder projects into navigable relationships.
+
+PartMember:
+    'part'
+    (bindingOp=':>>' name=MemberName ('=' boundRef=QualifiedName)?
+    | name=MemberName
+      (':' type=QualifiedName multiplicity=Multiplicity?)?
+      (bindingOp=('=' | ':>') boundRef=QualifiedName)?)
+    ('{' body+=UsageBodyMember* '}' | ';');
+
+// ─── Reference member ───────────────────────────────────────────────────────
+// e.g. ref subject : TraceableElement[*];
+
+RefMember:
+    'ref' name=ID ':' type=QualifiedName multiplicity=Multiplicity? ';';
+
+// ─── Doc comment ─────────────────────────────────────────────────────────────
+// Captured as a single terminal to prevent hidden ML_COMMENT from consuming it
+
+DocComment:
+    content=DOC_COMMENT;
+
+// ─── Specialization ──────────────────────────────────────────────────────────
+
+Specialization:
+    (':>' | 'specializes') superType=QualifiedName;
+
+// ─── Qualified name (data type rule → returns string) ────────────────────────
+
+QualifiedName returns string:
+    QualifiedNameSegment ('::' QualifiedNameTailSegment)*;
+
+QualifiedNameSegment returns string:
+    ID | 'library';
+
+// Non-first segments may be keywords so enum literals declared via
+// EnumLiteralName (e.g. DirectionKind::in, UseErrorCategoryKind::action)
+// are referenceable in attribute values. The first segment stays strict to
+// avoid ambiguity with usage prefixes at statement start.
+QualifiedNameTailSegment returns string:
+    QualifiedNameSegment
+    | 'in' | 'out' | 'inout' | 'part' | 'requirement' | 'action' | 'item' | 'port'
+    | 'interface' | 'connection' | 'attribute' | 'constraint' | 'enum' | 'view' | 'viewpoint'
+    | 'ref' | 'import' | 'package' | 'flow' | 'first' | 'allocate' | 'private' | 'public';
+
+// Allow keywords as identifiers in member/feature contexts
+MemberName returns string:
+    ID | 'requirement' | 'action' | 'item' | 'port' | 'interface' | 'connection'
+    | 'attribute' | 'constraint' | 'view' | 'viewpoint' | 'ref' | 'part';
+
+FeatureName returns string:
+    ID | 'part' | 'requirement' | 'action' | 'item' | 'port' | 'interface'
+    | 'connection' | 'flow' | 'view' | 'viewpoint' | 'attribute' | 'constraint'
+    | 'ref' | 'allocate' | 'end' | 'enum' | 'function' | 'operator';
+
+// ─── Hidden terminals (skipped by lexer) ─────────────────────────────────────
+
+hidden terminal WS: /\s+/;
+hidden terminal SL_COMMENT: /\/\/[^\n\r]*/;
+hidden terminal ML_COMMENT: /\/\*[\s\S]*?\*\//;
+
+// ─── Visible terminals ──────────────────────────────────────────────────────
+
+// DOC_COMMENT captures `doc /* ... */` as a single token.
+// The `doc\s+` prefix ensures it only matches when preceded by the `doc` keyword,
+// preventing conflict with ML_COMMENT for standalone /* ... */ comments.
+terminal DOC_COMMENT: /doc\s+\/\*[\s\S]*?\*\//;
+
+terminal ID: /[a-zA-Z_][a-zA-Z0-9_]*/;
+terminal STRING: /"(?:[^"\\]|\\.)*"/;
+// REAL must be declared before INT so the longest-match lexer emits a single
+// REAL token for `20.0` rather than INT `.` INT.
+terminal REAL: /[0-9]+\.[0-9]+/;
+terminal INT: /[0-9]+/;